In message <Pine.BSF.3.95q.980303225603.8985F-100000@misery.sdf.com>, Tom <tom@sdf.com> writes: | | | On Tue, 3 Mar 1998, Adam Shand wrote: | | > readers. if you were doing this with nntpcache the main advantage i would | > see wouldn't be reduction of charge from the news provider but rather | > improved response times for your clients and reduced bandwidth costs | > because of caching. | > | > you can't have your cake and eat it too ;) | | Hmm, the savings could still be substantional. There are a lot of users | at this site, but few seem to read news. According to innstats, about 50 | to 60K articles are read a day. Very wasteful since the server receives | 300 to 500K articles per day. NNTPCache can be extremely effecient, on a system with 7,000 users of which maybe 20% read news I get the following stats: NNTPCACHE 1.0.7.1 statistics on iris.nyx.net, DEFAULT:119 OVERALL: Statistics gathering commenced: 30 Jan 1998 21:16:16 MST Current nntpcached started: 03 Mar 1998 10:08:36 MST Servers run during statistical period: 15 Clients active: 7 Client connects: 43036 Outing server connections: 32464 Outing server connections (failed): 36 Total data from remote servers (bytes): 14.85G Total data to remote servers (bytes): 47.38M Total data from clients (bytes): 79.51M Total data to clients (bytes): 37.44G Total cache cache efficiency: 60.29% .... Database stores: 1116230 Database fetches: 614 Database size: 60.94M GROUP commands: 378231 LISTGROUP commands: 46381 Groups cached with > 0 articles: 14306 Groups expired: 0 Articles expired: 1532057 Posts: 3102 Posts (crossposted): 519 Posts (failed): 31 Posts (bytes): 7.47M Incoming crossposts: 1277694 Incoming crossposts (bytes): 18.38G LISTS: List-name---- --out-of-cache -----into-cache hits --xfer-per-hour lines -length active 669784 5.92G 1257 62.27M 99% 867.99 7.74M 20 750b active.times 2383 2.47G 62 29.41M 99% 3.16 3.23M 0 0b newsgroups 5799 2.50G 70 48.10M 98% 7.59 3.29M 0 0b overview_fmt 25865 1.99M 64 4.81k 100% 33.54 2.58k 0 0b MESSAGES: Data-type---- --out-of-cache -----into-cache hits ----------proxy xfer-per-hour article 34350 1.35G 137362 5.83G 19% 19203 35.56M 222 9.28M head 793481 750.87M 604220 547.74M 58% 70060 55.30M 1808 1.68M body 141077 1.56G 374649 6.40G 20% 39820 41.89M 667 10.29M xover 23544527 7.96G 3969695 1.40G 85% 23842 331.19M 35621 12.54M xhdr 0 0b 0 0b 100% 161340 2.19M 209 2.83k msgid 614 56.80k 1116230 99.97M 0% 1445 129.39k In short it gives us significant bandwidth and speed/cache savings. Paul. -- The Internet Business Ltd, Holywood House, Innis Court, Holywood, BT18 9HF pgregg@tibus.net http://www.tibus.net/ Phone: +44 (0)1232-424190 Fax: +44 (0)1232-424709 Eight out of every five people are math illiterates.
